Re: mpz_root assertion failure

[Jason Moxham]

On Saturday 30 Nov 2002 7:12 pm, Torbjorn Granlund wrote:
> Jason Moxham <address@hidden> writes:
>
>   gmp-4.1.1 appears to be OK
>
> Perhaps this patch is what made it work?
>
> *** mpn/generic/rootrem.c       8 May 2002 22:53:52 -0000       1.3
> --- mpn/generic/rootrem.c       11 May 2002 14:23:33 -0000      1.4
> ***************
> *** 86,88 ****
>
> !   qp = TMP_ALLOC_LIMBS (un);  /* should try and reduce this allocation */
>     xp = TMP_ALLOC_LIMBS (xn + 1);
> --- 86,88 ----
>
> !   qp = TMP_ALLOC_LIMBS (un + 1);
>     xp = TMP_ALLOC_LIMBS (xn + 1);


NO

but

I replaced 
gmp-4.1/mpn/generic/rootrem.c
with
gmp-4.1.1/mpn/generic/rootrem.c

and all is OK

the only difference between these files is

diff gmp-4.1.1/mpn/generic/rootrem.c gmp-4.1/mpn/generic/rootrem.c

73c73
<   xnb = (unb - 1) / nth + 1;
---
>   xnb = (unb + nth - 1) / nth;

which is strange as the test case has only 114098 decimals ~3.6e6 bits so we 
cant be overflowing a long type , so it strongly  suggests a gcc error